LPR Processing Units running SharpOS 12.5 or higher use machine learning to classify license plates and to perform character recognition.
Although off-the-shelf machine learning solutions are commercially available for LPR, their results can be unpredictable. With our hardware and software engineering expertise, we have built the AutoVu™ MLC from the ground up. This includes developing a deep neural network (DNN), training the system with Sharp camera images, and optimizing the system to run on existing LPR Processing Unit hardware. As a result, depending on the regional contexts, you can expect up to a 50% reduction in plate capture errors and character recognition errors when compared to widely used classical algorithms. The AutoVu™ MLC represents a significant reduction in the time that operators spend manually correcting plate reads.
As the AutoVu™ MLC continues to develop, we will apply it to more aspects of the LPR process. You will be able to take advantage of these improvements with future SharpOS releases.